Instructions to use danielrosehill/Contextual-Sentence-Creator-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use danielrosehill/Contextual-Sentence-Creator-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf danielrosehill/Contextual-Sentence-Creator-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf danielrosehill/Contextual-Sentence-Creator-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf danielrosehill/Contextual-Sentence-Creator-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf danielrosehill/Contextual-Sentence-Creator-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf danielrosehill/Contextual-Sentence-Creator-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf danielrosehill/Contextual-Sentence-Creator-GGUF:Q4_K_M

Use Docker

docker model run hf.co/danielrosehill/Contextual-Sentence-Creator-GGUF:Q4_K_M

Llama new
LM Studio
Jan

vLLM

How to use danielrosehill/Contextual-Sentence-Creator-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "danielrosehill/Contextual-Sentence-Creator-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "danielrosehill/Contextual-Sentence-Creator-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/danielrosehill/Contextual-Sentence-Creator-GGUF:Q4_K_M

Ollama
How to use danielrosehill/Contextual-Sentence-Creator-GGUF with Ollama:
```
ollama run hf.co/danielrosehill/Contextual-Sentence-Creator-GGUF:Q4_K_M
```
Unsloth Desktop
Docker Model Runner
How to use danielrosehill/Contextual-Sentence-Creator-GGUF with Docker Model Runner:
```
docker model run hf.co/danielrosehill/Contextual-Sentence-Creator-GGUF:Q4_K_M
```

Lemonade

How to use danielrosehill/Contextual-Sentence-Creator-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull danielrosehill/Contextual-Sentence-Creator-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Contextual-Sentence-Creator-GGUF-Q4_K_M

List all available models

lemonade list

Atomic Chat

Term-Context Sentence Generator (Granite 4.2 3B, v1), GGUF

GGUF builds of danielrosehill/Contextual-Sentence-Creator. Prompt format, evaluation and limitations are on that page.

file	size	note
`term-context-sentence-generator-granite-4.2-3b-v1-Q4_K_M.gguf`	2.2 GB	default for CPU
`term-context-sentence-generator-granite-4.2-3b-v1-Q8_0.gguf`	3.9 GB	closest to the bf16 weights

The chat template in these files is replaced by the exact one the model was trained with (chat_template.jinja): ChatML ending in <|im_start|>assistant\n<think></think>, with thinking off. Granite's stock template leaves <think> open, and the model then answers in free text instead of a JSON array. If your runtime overrides the embedded template, pass chat_template.jinja.

llama-server -m term-context-sentence-generator-granite-4.2-3b-v1-Q4_K_M.gguf -c 4096
curl -s localhost:8080/v1/chat/completions -H 'content-type: application/json' -d '{
 "messages": [
  {"role": "system", "content": "Write exactly n different natural English sentences, 6 to 18 words each, that each use the term once, exactly as given, in the meaning the context describes. Reply with a JSON array of strings."},
  {"role": "user", "content": "{\"term\": \"gemach\", \"context\": \"a free community lending fund\", \"n\": 10}"}],
 "temperature": 0.8, "top_p": 0.95, "max_tokens": 640}'

Measured 2026-11-03 on a laptop CPU (8 threads, Q4_K_M, llama.cpp build of that day): about 12.5 tokens/s, 12 to 16 s for 10 sentences. 4/4 test calls returned valid JSON with exactly 10 sentences.

Converted with llama.cpp convert_hf_to_gguf.py (F16) and then llama-quantize. The template was set with gguf_new_metadata.py --chat-template.